Robotics becomes magnet for money
By Charles Sheehan
PITTSBURGH Researchers in robotics have traditionally faced two debilitating obstacles: terribly expensive parts and difficulty attracting financing from anyone outside of a small corps of true believers.
But the field could be in line for a major jolt. Robotics experts see a "perfect storm" heading their way, thanks in no small part to the human ravages of war.
Just as the constant march of technology is driving down the cost of key components, top universities in robotics are reporting major increases in federal financing, with the Defense Department the biggest spender.
The military desperately wants to reduce the number of soldiers killed by roadside bombs or surface-to-air missiles cheap implements of war that have felled scores in Iraq. Many in the Pentagon believe the answer lies in autonomous air, sea and land vehicles.
The Robotics Institute at Carnegie Mellon University has seen federal financing jump 48 percent since 2000, and by 117 percent since 1994. Much of the $24.8 million in federal money for 2003 came from the Pentagon, said institute director Chuck Thorpe.
The university's corporate financing for robotics is also up 40 percent since 2000, with $7.8 million arriving last year.
Other universities, such as the California, Virginia and Georgia institutes of technology, say money for robotics is up at least 50 percent or more in recent years.
At the same time, the materials that comprise the most technologically advanced components in robots, from optics to software, are becoming "dirt cheap," said Dan Kara, who covers the industry for Robotics Trends.
Technology that lets robots perceive and overcome obstacles has made unparalleled bounds largely because the cost of charge-coupled devices, microprocessors and varied sensors has fallen away as rapidly as computing power and memory have expanded.
"Nobody is inventing the wheel anymore," Kara said. "The core of research that occurred over the last 10 years is driving this market intellectually and now there's a ton of money coming from the military side of the aisle."
The Pentagon, which spent $3 billion on unmanned aerial vehicles between 1991 and 1999, is expected to spend upward of $10 billion through 2010. Under a congressional mandate, the Defense Department is pushing for one-third of its ground vehicles to be unmanned by 2015.
The Army is seeking portable reconnaissance robots, transport robots and fighting vehicles that could be deployed in place of the Abrams tank. The 42-pound PackBot, which can climb stairs and work under water, has been used by U.S. troops flushing out Afghan caves. The Marines have developed a similar robot half that size.
The Pentagon's research arm, the Defense Advanced Research Projects Agency, is sponsoring more than 40 projects in robotics, spokeswoman Jan Walker said.
CalTech's Joel Burdick, an engineering professor who has been involved in robotics since the early 1980s, is using DARPA money to develop neuroprosthetics with which the brain could control machinery, "giving pilots a third hand, so to speak," he said.
"Funding is largely military now," Burdick said. "The projects are more focused on applications. It's not seed money."
It is difficult to determine exactly how much money is going into research at universities because government contracts also go to corporations, such as Boeing Co., that work closely with schools like Carnegie Mellon.
The engineers and computer scientists behind the machines have even found themselves in the national spotlight. Last month, many were mobbed by camera crews in California's Mojave Desert as 15 teams lined up for an autonomous 150-mile race. The prize: $1 million from DARPA.
Although no team won the money, DARPA's Grand Challenge is just one of several competitions that will draw top robotics talent from across the country this year.
Robert Michelson, a principal research engineer at Georgia Tech, is holding the fourth annual International Aerial Robotics Competition in July.
Robotic aircraft will be required to fly 1.8 miles to an urban setting, find a building, then enter it via a window or a hole in the roof to find a target inside. The robot must then transmit an image back to base all without human interference.
"This is not a paper exercise," Michelson said. "This is real-world show-me."